1. Field of the Invention
The present invention relates generally to semiconductors and, more particularly, to forming circuitry using wafer bonding.
2. Description of the Related Art
Advances in semiconductor manufacturing technology have provided computer systems with integrated circuits that include many millions of active and passive electronic devices, along with the interconnects to provide the desired circuit connections. A typical computer system includes a computer chip, with processor and control circuits, and an external memory chip. As is well-known, most integrated circuits include laterally oriented active and passive electronic devices that are carried on a single major surface of a substrate. The current flow through laterally oriented devices is generally parallel to the single major surface of the substrate. Active devices typically include transistors and passive devices typically include resistors, capacitors, and inductors. However, these laterally oriented devices consume significant amounts of chip area. Sometimes laterally oriented devices are referred to as planar or horizontal devices. Examples of laterally oriented devices can be found in U.S. Pat. No. 6,600,173 to Tiwari, U.S. Pat. No. 6,222,251 to Holloway and U.S. Pat. No. 6,331,468 to Aronowitz.
Vertically oriented devices extend in a direction that is generally perpendicular to the single major surface of the substrate. The current flow through vertically oriented devices is generally perpendicular to the single major surface of the substrate. Hence, the current flow through a vertically oriented semiconductor device is generally perpendicular to the current flow through a horizontally oriented semiconductor device. Examples of vertically oriented semiconductor device can be found in U.S. Pat. No. 5,106,775 to Kaga, U.S. Pat. No. 6,229,161 to Nemati, U.S. Pat. No. 7,078,739 to Nemati. It should be noted that U.S. Pat. No. 5,554,870 to Fitch, U.S. Pat. No. 6,229,161 to Nemati and U.S. Pat. No. 7,078,739 to Nemati disclose the formation of both horizontal and vertical semiconductor devices on a single major surface of a substrate.
It is desirable to provide computer chips that can operate faster so that they can process more data in a given amount of time. The speed of operation of a computer chip is typically measured in the number of instructions in a given amount of time it can perform. Computer chips can be made to process more data in a given amount of time in several ways. For example, they can be made faster by decreasing the time it takes to perform certain tasks, such as storing and retrieving information to and from the memory chip. The time needed to store and retrieve information to and from the memory chip can be decreased by embedding the memory devices included therein with the computer chip. This can be done by positioning the memory devices on the same surface as the other devices carried by the substrate.
However, there are several problems with doing this. One problem is that the masks used to fabricate the memory devices are generally not compatible with the masks used to fabricate the other devices on the computer chip. Hence, it is more complex and expensive to fabricate a computer chip with memory embedded in this way. Another problem is that memory devices tend to be large and occupy a significant amount of area. Hence, if most of the area on the computer chip is occupied by memory devices, then there is less area for the other devices. Further, the yield of the computer chips fabricated in a run decreases as their area increases, which increases the overall cost.
Instead of embedding the memory devices on the same surface as the other devices, the memory chip can be bonded to the computer chip to form a stack, as in a 3-D package or a 3-D integrated circuit (IC). Conventional 3-D packages and 3-D ICs both include a substrate with a memory circuit bonded to it by a bonding region positioned therebetween. The memory chip typically includes lateral memory devices which are prefabricated before the bonding takes place. In both the 3-D package and 3-D ICs, the memory and computer chips include large bonding pads coupled to their respective circuits. However, in the 3-D package, the bonding pads are connected together using wire bonds so that the memory and computer chips can communicate with each other. In the 3-D IC, the bonding pads are connected together using high pitch conductive interconnects which extend therebetween. Examples of 3-D ICs are disclosed in U.S. Pat. Nos. 5,087,585, 5,308,782, 5,355,022, 5,915,167, 5,998,808 and 6,943,067.
There are several problems, however, with using 3-D packages and 3-D ICs. One problem is that the use of wire bonds increases the access time between the computer and memory chips because the impedance of wire bonds and large contact pads is high. The contact pads are large in 3-D packages to make it easier to attach the wire bonds thereto. Similarly, the contact pads in 3-D ICs have correspondingly large capacitances which also increase the access time between the processor and memory circuits. The contact pads are large in 3-D ICs to make the alignment between the computer and memory chips easier. These chips need to be properly aligned with each other and the interconnects because the memory devices carried by the memory chip and the electronic devices carried by the computer chip are prefabricated before the bonding takes place.
Another problem with using 3-D packages and 3-D ICs is cost. The use of wire bonds is expensive because it is difficult to attach them between the processor and memory circuits and requires expensive equipment. Further, it requires expensive equipment to align the various devices in the 3-D IC. The bonding and alignment is made even more difficult and expensive because of the trend to scale devices to smaller dimensions. It is also very difficult to fabricate high pitch conductive interconnects.
Accordingly, it is highly desirable to provide a new method for embedding memory devices using wafer bonding which is cost effective and reliable.